Submitted by Andreas Rössler.

Submission data

Full nameRahmouni c40
DescriptionSee FaceForensics ++ paper. A classifier based on the work of Rahmouni et al. trained on the c40 compressed data.
Publication titleDistinguishing computer graphics from natural images using convolution neural networks
Publication authorsNicolas Rahmouni, Vincent Nozick, Junichi Yamagishi, and Isao Echizen
Publication venueIEEE Workshop on Information Forensics and Security,
Publication URLhttp://www-igm.univ-mlv.fr/~vnozick/publications/Rahmouni_WIFS_2017/Rahmouni_WIFS_2017.pdf
Input Data TypesUses Face Detection
Programming language(s)Python, Keras
HardwareIntel Xeon 6140, 2080ti
Submission creation date29 Aug, 2019
Last edited29 Aug, 2019

Binary Classification v3

InfoDeepfakesFace2FaceFaceSwapNeuralTexturesPristineTotal
0.8550.6420.5630.6070.5000.581